Modern, automated factories involve the use of "robots" in performance of required and often repetitive tasks in the manufacturing process. Such manufacturing robots need to "see" the objects they are working with in order to perform their programmed functions, sense and avoid foreign objects such as people entering their workspace, etc. Robots which are autonomous vehicles must detect and avoid path obstructions and remotely piloted vehicles are required to provide their operators with 3-D data about their surroundings.
Several systems have been suggested to perform these functions. Video parallax systems use two video cameras to record the same scene from different angles. A computer then matches points from the two pictures and calculates the range from the angle between them. However, this computer matching requires sophisticated edge detection and pattern recognition algorithms. As a result, systems based on this technique tend to be slow or very expensive.
Another method of extracting 3-D information involves the transmission of radar or ultrasonic pulses. However, these systems lack sufficient angular resolution necessary for many functions. These systems also often require mechanical aiming of either the transmit or receive transducer which reduces reliability further.